The link above has some recommendations, like making sure that both FI will have the same software and same amount of server and uplink ports, etc...
2-Adding a chassis is not disruptive, just be careful if you are going to add additional uplinks while using dynamic pinnning and not static pinning. If you have dynamic pinning (which is the default) and you add more uplinks ports, the server-uplink pinning will be "recalculated" for " load balancing" among uplinks.
Bottom line, if you will just add more server ports and the amount of uplink ports will be exactly the same, there should be no disruption.
Let us know if there are more doubts before attempting the changes.
1). I have actually been using this document for my project, and I think I worked out a pretty good migration plan based on it, but there are some grey areas/questions, MARKED IN BOLD CAPS. Could you clarify them for me?
Step 1 Upgrade UCS firmware across all the devices to the same version - 2.1(1f). THE NEW 6248S CAN BE TEMPORARY CONFIGURED AS STANDALONE OR IN A CLUSTER FOR THE FIRMWARE UPGRADE, AND THEN THE CONFIGURATION SHOULD BE WIPED OUT AND THE FABRIC INTERCONNECTS POWERED OFF.
Step 2 Using either the UCS manager CLI or GUI, verify the state (subordinate or active) of the existing (production) fabric interconnects.
!!! Step 3 Back up the software configuration information and the UCS Manager software !!!
Step 4 Disable the server ports on the subordinate fabric interconnect – NEW or EXISTING? THIS STEP IS CONFUSING AND IT IS REFERENCED IN STEP 17.
Step 5 Power down the EXISTING subordinate fabric interconnect by unplugging it from the power source. If you are monitoring the upgrade using a KVM session, you may need to reconnect the KVM session when you power down the fabric interconnect.
THE NEW FABRIC INTERCONNECT(S) SHOULD BE POWERED OFF AT THIS POINT
Step 6 Disconnect the cables running from the chassis IO Modules to the subordinate fabric interconnect ports in slot 1 (the old EXISTING fabric interconnect).
Step 7 Connect these cables into the corresponding ports in slot 1 on one of the new Cisco UCS 6248 UP fabric interconnect, using the connection records to preserve the port mapping and the configured server pinning.
See Fabric Interconnect Port Connection Record, on page 7.
See Fabric Interconnect Upgrade Considerations, on page 1.
Step 8 Disconnect the L1/L2, M1 management, and Console cables on the old fabric interconnect. The ports for these connections are on the opposite side of the interconnect, so if your cables are just barely long enough to connect two rack-adjacent UCS 6120 interconnects you will probably need new cables.
Step 9 Connect the M1 management, and Console cables to the new Cisco UCS 6248 UP.
Step 10 Connect the L1/L2 cables that were disconnected to the new Cisco UCS 6248 UP. L1 connects to L1, L2 connects to L2.
Step 11 Disconnect the Ethernet or FC cables from slot 2 of the old fabric interconnect.
Step 12 Connect the Ethernet or FC cables to the corresponding ports in slot 2 of the new Cisco UCS 6248 UP. Some may go to slot 1, depending on the mappings planned out earlier in the process
I HAVE NO SLOT 2 ON THE NEW FABRIC INTERCONNECTS ON THE OLD ONES, SIX PORTS IN SLOT 2 ARE CONFIGURED AS FIBER CHANNEL PORTS. WILL I NEED TO MANUALLY CONFIGURE LAST SIX PORTS AS FC PORTS? IF YES, AT WHAT STEP IS THIS CONFIGURATION DONE, AND ARE THE FIBER CHANNEL CABLES SUPPOSED TO BE PLUGGED IN BEFORE OR AFTER THE PORTS ARE CONFIGURED AS FC PORTS?
Step 13 Connect the power to the new Cisco UCS 6248 UP, it will automatically boot and run POST tests.
Important: Directly connect the console cable to a terminal and observe the boot sequence. You should at some point see the Basic System Configuration Dialog, where you will configure the switch as a subordinate interconnect. - IS THIS THE WIZARD THAT RUNS ON EVERY UNCONFIGURED FABRIC INTERCONNECT WHEN YOU POWER IT ON, WHEN YOU HAVE TO GIVE IP, NAME, DNS, ETC.? CAN I USE THE SAME IP THAT WAS ON THE OLD FABRIC INTERCONNECT?
If you do not see this dialog, you either have different builds of software on your old primary and new subordinate, or the new subordinate has previously been part of a cluster and will need to have all configuration information wiped before it can be added to a cluster as a subordinate. In either case, immediately disconnect the L1 and L2 connections and complete the bringup as a standalone fabric interconnect, then correct the issue before proceeding further.
I AM NOT SURE IF THE NEXT STEP IS APPLICABLE TO ME AT ALL, AS I DO NOT HAVE N10-E0440, I HAVE N10-E0060. COULD YOU COMMENT ON THIS?
Step 14 (Optional) Remap UCS 6100 fabric interconnect FC ports 2/1 to 2/4 on a N10-E0440 expansion module or any slot 3 ports onto the new fabric interconnect expansion module.
a) Use UCS Manager to delete the ports on the subordinate fabric interconnect that you will need to move within the configuration - WHERER DO I DELETE THE PORTS - ON THE NEW ONE?
b) For each port you have just deleted, create new ports on either slot 1 or slot 2. These ports must use the same port type definitions as the old ports, but will use different port numbers.
c) For recently moved Ethernet server ports, reconfigure the associated service profile to use the new port number for the appropriate LAN pin group.
d) For recently moved uplink Ethernet ports, reconfigure the port channel settings to use the new ports.
e) For recently moved uplink FC ports, reconfigure the associated service profile SAN pin group to use the new ports.
f) Re-acknowledge chassis for blade servers and fabric extender for rack servers.
This will be disruptive to traffic, but is necessary in this specific scenario. DO I LOOSE CONNECTIVITY FOR THE ENTIRE CHASSIS, OR JUST ONE FABRIC?
Step 15 The new subordinate fabric interconnect will automatically synchronize the configuration and database/state information from the primary fabric interconnect. Synchronization between primary and subordinate fabric interconnects can take several minutes. The port configuration is copied from the subordinate switch to the new hardware.
Step 16 Verify that the data path is ready.
See Verifying that the Data Path is Ready.
Make sure all faults are resolved before proceeding.
a) Verify and if necessary reconfigure the SAN pin group for FC ports in the associated service profile.
b) Verify and if necessary reconfigure the LAN pin group for Ethernet ports in the associated service profile.
c) Verify and if necessary reconfigure the port channel for uplink Ethernet ports.
THE FOLLOWING STEP IS ALSO CONFUSING. IF IT IS TALKING ABOUT THE OLD SUBORDINATE FABRIC INTERCONNECT THAT IS BEING REPLACED, WHY DO WE NEED TO ENABLE THE SERVER PORTS ON IT, THERE IS NOTHING CONNECTED TO IT. IF IT IS TALKING ABOUT THE NEW ONE - 6248, THEN, IT MEANS THEY HAVE TO BE DISABLED IN STEP 4, BUT IT MAKES NO SENSE, BECAUSE THEY ARE NOT SUPPOSED TO HAVE ANY CONFIGURATION ON IT. CAN YOU CLARIFY?
Step 17 Enable the server ports that had been disabled in Step 4. -
a) If you have changed port mappings, you may need to re-acknowledge the chassis or rack server connected to the subordinate fabric interconnect. DOES IT APPLY TO ALL TYPES OF PORTS – FC, SERVER, UPLINK ETHERNET PORTS?
b) Verify and if necessary reconfigure Ethernet ports as server ports.
Step 18 Promote the subordinate fabric interconnect to active, and repeat the process on the second Cisco UCS 6248 UP. Cable the second new fabric interconnect identically to the first, and allow the reconfiguration done to be applied to the second new fabric interconnect as well. DOES THE SECOND NEW 6248 NEED TO BE OFF OR ON WHEN CABLING IT? DO I NEED TO GO THROUGH THE WIZARD AT THE POWER ON TO MAKE IT A PART OF THE CLUSTER?
See Forcing a Fabric Interconnect Failover.
2). Does CISCO have a document outlining the steps for adding a chassis?
Step-1 <<< For what I understand you will only do a HW upgrade and not a SW upgrade, if I am correct, all you have to do is make sure that both FIs run the same firmware before adding the new FI to the cluster, that is to avoid inconsistencies that might cause unexpected behaviors.
The idea is to first replace the subordinate FI to have minimum disruption and then proceed with the other one. <<< I think this also answers the question in regards to step 4 & 5
Step-12 <<< That is in case you have an expansion module in the first FI, here the idea again is to keep the same cabling
Step-13 <<<< You should be able to use the same IP address, is like when you are replacing a computer for another one, the IP is reusable as far as no one else in the network has the same IP
Step-14 <<< It actually says optional
Step-16 <<< We need to remember that you disable the ports in step 4, and now those cables are in the new FI (disabled)
Step 18 <<< You need to go through the wizard, it will tell you that it detected a peer and will ask you if you want to join the cluster, if yes, it will ask you for the admin password and the rest will be automatic
2). Does CISCO have a document outlining the steps for adding a chassis?
I am personally not aware of a document like that one (not that it does not exist, I just have not seen it), but the process is pretty simple.
Make sure that the quantity of links going from each IOM to each FI meets the amount of links specified in the chassis discovery policy in UCSM and that each IOM goes to only one FI.
I see you are a partner, hence, I would like to take advantage and invite you to open a case with our PDI (Planning, Design and Implementation) team, they can certainly help you with these questions and help you all the way through the process of this HW upgrade:
Usually, we can access ESXi Shell by pressing Alt+F1 from ESXi DCUI (Direct Console User Interface).
But on HyperFlex system, it just shows black window.
This is expected behavior because HyperFlex redirects ESXi Shell output to SoL...
Configuring an Export Policy Using the GUI
This procedure explains how to configure an Export policy using the APIC GUI. Follow these steps to trigger a backup of your data:
On the menu bar, choose Admi...
RBAC users like firstname.lastname@example.org may fail HX Connect login. At that time, "Incorrect user name or password(100005)" is shown as a failure reason.
RBAC users can login to vCenter server. So, RBAC username and passwo...